Merge pull request scikit-learn#1656 from rlmv/idf_diag

Fix AttributeError caused by an unlearned idf vector

Author: GaelVaroquaux

sklearn/feature_extraction/tests/test_text.py

@@ -389,6 +389,10 @@ def test_vectorizer():
     t2 = TfidfTransformer(norm='l1', use_idf=False)
     tf = t2.fit(counts_train).transform(counts_train).toarray()
     assert_equal(t2.idf_, None)
+    
+    # test idf transform with unlearned idf vector
+    t3 = TfidfTransformer(use_idf=True)
+    assert_raises(ValueError, t3.transform, counts_train)
 
     # L1-normalized term frequencies sum to one
     assert_array_almost_equal(np.sum(tf, axis=1), [1.0] * n_train)

sklearn/feature_extraction/text.py

@@ -915,6 +915,8 @@ def transform(self, X, copy=True):
             X.data += 1
 
         if self.use_idf:
+            if not hasattr(self, "_idf_diag"):
+                raise ValueError("idf vector not fitted")  
             expected_n_features = self._idf_diag.shape[0]
             if n_features != expected_n_features:
                 raise ValueError("Input has n_features=%d while the model"